Polaritons arise when photons couple strongly to vibrations of the crystal lattice, forming so-called phonon polaritons. In certain van der Waals materials, these waves propagate hyperbolically, meaning their wavefronts sweep outward in open arcs rather than closed ellipses, a property that allows light to be confined and steered with extraordinary precision. Alpha-MoO3 has become a star of this field because its in-plane optical response is naturally anisotropic: light propagates differently along different crystallographic directions, producing hyperbolic dispersion in the mid-infrared range. Over the past several years, researchers have twisted stacked flakes of such crystals against one another to create topological transitions in polariton propagation, discovering everything from photonic magic angles to ghost hyperbolic surface waves. But all of these approaches share a fundamental limitation: the optical response of a natural material is fixed by its crystal symmetry, so reconfiguration has typically required physically restacking, rotating, or chemically modifying the sample.

Conventional strategies for controlling polaritons rely on refractive-index engineering. By placing a polariton-bearing crystal on top of another optical medium, engineers can coax the waves in the two layers to hybridize, blending their dispersion relations into something new. Yet this coherent hybridization demands strict wavevector matching between the coupled modes, a condition that is difficult to satisfy and sensitive to sample geometry, thickness, and frequency. It also ties the resulting behavior inseparably to the passive optical properties of the underlying layer. The team behind the new study, led by researchers at Shanghai Jiao Tong University and the National Center for Nanoscience and Technology, asked a different question: what if, instead of matching wavevectors, one simply filtered out the waves one did not want?

The answer came in the form of an unlikely partner for the molybdenum oxide flake: a film of aligned carbon nanotubes. These dense arrays of nanotubes, all pointing in the same direction, are electrically conductive along their axis but effectively insulating across it, making them a naturally anisotropic conductor. Crucially, the researchers found that the carrier dynamics within the nanotube film are overdamped, meaning that charge carriers lose their coherent momentum too quickly to sustain any resonant optical response of their own. As a result, the nanotubes refuse to hybridize with the polaritons in the overlying crystal. Instead of coherently mixing, the two layers interact through a purely resistive, proximity-induced coupling—essentially, the polaritons leaking into the nanotube film are simply dissipated as heat.

That dissipation, however, is anything but random. Because the nanotubes conduct only along their alignment axis, they absorb polariton energy selectively depending on the direction in which the wave propagates relative to the nanotube axis. The aligned carbon nanotube layer thus acts as what the researchers call a momentum-space loss filter: in the reciprocal space that describes the polariton dispersion, wavevectors aligned with the nanotube axis are strongly attenuated while others survive. The consequence in real space is dramatic. The intrinsic hyperbolic dispersion of the alpha-MoO3 flake, with its elegant pairs of symmetric arcs, is reshaped into a symmetry-broken dispersion in which the energy flow—the Poynting vector—is inhibited in some directions and concentrated in others. Full-wave electromagnetic simulations confirmed that an effective medium description of the heterostructure reproduces the observed field patterns, with the asymmetry gradually diminishing as the crystal flake becomes thicker and the resistive coupling weaker.

The experimental evidence came from scanning near-field optical microscopy, a technique that drags a nanoscale antenna across the sample surface to image polariton fringes with resolution far below the diffraction limit. By preparing heterostructures in which the crystal’s [100] axis was rotated at various angles relative to the nanotube axis—from zero degrees through 90 degrees—the team observed a smooth, continuous transformation of the polariton wavefronts. At certain twist angles the familiar symmetric hyperbolas remained essentially intact, while at others the fringes warped into highly asymmetric, shear-like patterns reminiscent of the hyperbolic shear polaritons previously seen only in exotic low-symmetry crystals such as beta-gallium oxide or monoclinic semiconductors. Here, however, the shear was not baked into the crystal lattice; it was imposed by an external, deliberately engineered dissipative layer whose orientation could be chosen freely at fabrication time.

The most striking result is that the entire transformation is electrically reversible. Because the dissipative loss in the nanotube film is governed by its Drude response—the same free-carrier absorption that limits the conductivity of any metal—applying a gate voltage changes the carrier density in the nanotubes and thereby tunes the strength of the momentum-space filter. In the experiments, sweeping the gate voltage from positive to negative values continuously steered the polariton propagation from symmetric hyperbolic wavefronts to strongly sheared, asymmetric ones and back again. Unlike approaches that require physically altering the sample, this means the polariton topology can be reprogrammed in real time, with a single device cycling through a whole family of optical responses that a static crystal could never display.

The work also carries conceptual weight for the broader field of non-Hermitian physics. In standard quantum and optical systems, the Hamiltonian describing the dynamics is Hermitian, guaranteeing energy conservation; loss must then be treated as a nuisance or, at best, compensated with gain. Physicists have increasingly recognized that deliberately engineered loss—described by non-Hermitian formalisms—can be a resource in its own right, enabling exceptional points, parity-time symmetry effects, and non-Hermitian skin phenomena. The new study extends that philosophy into topological nanophotonics in a particularly clean way: rather than balancing gain and loss or exploiting fine-tuned couplings, the researchers show that a simple, gate-tunable anisotropic resistor is enough to break symmetry and reshape topology, establishing dissipation itself as a programmable degree of freedom in device design.

The implications reach across several technologies. Mid-infrared polaritons are already being explored for molecular sensing, thermal management, and sub-diffraction imaging, and recent demonstrations of hyperbolic electroluminescence suggest routes toward electrically driven polariton sources. A platform in which the directionality, asymmetry, and topology of those waves can be switched by a voltage opens the door to dynamically reconfigurable nanophotonic circuits, electrically steered thermal emitters, modulators, and beam-shaping elements operating at wavelengths far below what conventional optics allows. The aligned carbon nanotube films used in the study are compatible with wafer-scale growth and high-performance electronics, hinting that the integration of polaritonic and electronic functionality on a single chip may be practical rather than aspirational. More broadly, the study suggests that the road to programmable nanophotonics may run not only through what materials let light do, but through what they quietly prevent it from doing—a reminder that in modern optics, sometimes the most powerful design tool is controlled loss.

Subject of Research: Electrically programmable polariton symmetry breaking via non-Hermitian dissipation engineering in van der Waals heterostructures

Article Title: Electrically programmable polariton symmetry breaking via non-Hermitian dissipation engineering

Article References: Chen, N., Teng, H., Sun, Y., Yang, Y., Xue, Z., You, O., Chen, K., Jiang, C., Wang, J., Zhou, S., Liu, X., Wang, C., Li, Z.-Z., Meng, S., Zhu, M., Hu, H., & Dai, Q. (2026). Electrically programmable polariton symmetry breaking via non-Hermitian dissipation engineering. Nature Materials. https://doi.org/10.1038/s41563-026-02734-z

Image Credits: AI Generated

DOI: 10.1038/s41563-026-02734-z

Keywords: phonon polaritons, alpha-MoO3, aligned carbon nanotubes, non-Hermitian photonics, dissipation engineering, symmetry breaking, hyperbolic dispersion, shear polaritons, electrical gating, nanophotonics, van der Waals heterostructures, topological photonics